amplify 40kHz audio signal using TL082: first two stages are fine, but high noise from the third sta

[for_idea]

Dear friends,

I am making an ultrasound signal receiver. The transmitter and receiver
is apart from each other about 6 meters. The circuit is powered by a 9v
battery. I used three amplifiers from two TL082. The reference voltage
(about 4.5v) is generated from voltage divider (two 100k resisters in
series). All amplifiers are in inverting input mode. First and second
stages are configured as: 10k input resister and 500k feedback
resister. The signal output in the second stage is very good. However,
the signal from the third stage (input res.= 10k, output res. = 200k)
is significantly corrupted by noise. Please give me some advice to
clean up the amplified signal.

Thanks,

Zhi Yang

 

[Larry Brasfield]

Dear friends,

I am making an ultrasound signal receiver. The transmitter and receiver
is apart from each other about 6 meters. The circuit is powered by a 9v
battery. I used three amplifiers from two TL082. The reference voltage
(about 4.5v) is generated from voltage divider (two 100k resisters in
series). All amplifiers are in inverting input mode. First and second
stages are configured as: 10k input resister and 500k feedback
resister. The signal output in the second stage is very good. However,
the signal from the third stage (input res.= 10k, output res. = 200k)
is significantly corrupted by noise. Please give me some advice to
clean up the amplified signal.

If my stated assumptions are incorrect, you can
take them as a form of advice.

I assume your single-supply circuit is referenced to
a "pseudo ground" developed by that divider.

I assume the amplifier supplies are well bypassed
to that pseudo ground at the frequencies you care
about, (and beyond for stability's sake).

I assume that your amplifiers stages are the simple,
Rfeedback/Rin configuration and nothing really
boneheaded is hidden by the invisible schematic.

I assume that you would not say "noise" when you
meant "oscillation".

I assume that you would not say "noise" when you
mean "interference getting into the input because of
inadequate shielding or shield connection".

I assume that you know about frequency selective
filtering to remove noise from those parts of the
spectrum you do not intend to capture as signal.

Now, given all that, you should expect some noise,
and you should expect the most in the last gain stage.
So I do not see what can be done unless the noise
is excessive, and even then I would need to see a
schematic. See http://www.tech-chat.de/aacircuit.html
for a tool enabling you to post your schematic here.

Thanks,

You're welcome.

 

[Larry Brasfield]

Slight correction/amendment inserted below.

If my stated assumptions are incorrect, you can
take them as a form of advice.

I assume your single-supply circuit is referenced to
a "pseudo ground" developed by that divider.

I assume the amplifier supplies are well bypassed
to that pseudo ground at the frequencies you care
about, (and beyond for stability's sake).

I assume that your amplifiers stages are the simple,
Rfeedback/Rin configuration and nothing really
boneheaded is hidden by the invisible schematic.

I assume that you would not say "noise" when you
meant "oscillation".

I assume that you would not say "noise" when you
mean "interference getting into the input because of
inadequate shielding or shield connection".

I assume that you know about frequency selective
filtering to remove noise from those parts of the
spectrum you do not intend to capture as signal.

I assume that the choice of TL082 is appropriate
for the source impedance of your transducer. If
you are not sure of this, you should mention what
it is and what you know about that impedance.

I assume that your choice of a 10k input resistor
is made with knowledge of thermal noise, and is
not the source of input referred noise higher than
your application can tolerate. This again depends
on the transducer characteristics.

 

[John Woodgate]

(in <[email protected]>) about
'amplify 40kHz audio signal using TL082: first two stages are fine, but
high noise from the third stage', on Wed, 16 Mar 2005:

Dear friends,

I am making an ultrasound signal receiver.

For what frequency?

The transmitter and receiver
is apart from each other about 6 meters. The circuit is powered by a 9v
battery. I used three amplifiers from two TL082. The reference voltage
(about 4.5v) is generated from voltage divider (two 100k resisters in
series). All amplifiers are in inverting input mode. First and second
stages are configured as: 10k input resister and 500k feedback
resister.

50 times gain from a TL082 at an'ultrasonic' frequency is pushing its
capability. It MAY be OK, but a faster op-amp would be better.

The signal output in the second stage is very good.

What is the peak-to-peak amplitude? With +/- 4.5 V supplies, you can
only get about 8 V peak-to-peak.

However,
the signal from the third stage (input res.= 10k, output res. = 200k)
is significantly corrupted by noise. Please give me some advice to
clean up the amplified signal.

How are you coupling the stages together? If you have DC-coupled them
you are amplifying the first op-amp's input offset by 50 x 50 x 20 = a
lot. 100 uV of offset at the input becomes 5 V at the output - enough to
seriously compromise your results.

 

[Joerg]

Hello John,

50 times gain from a TL082 at an'ultrasonic' frequency is pushing its
capability. It MAY be OK, but a faster op-amp would be better.

He said 40kHz. But you are right, 50 times is pushing it for a 3MHz
unity gain BW opamp. And per datasheet that's only the typical value.
Quite frankly, for a 40kHz ultrasound amp I'd do it with plain transistors.

What is the peak-to-peak amplitude? With +/- 4.5 V supplies, you can
only get about 8 V peak-to-peak.

If you are lucky and there is literally no load.

Regards, Joerg

 

[for_idea]

Dear Larry,

Thanks for your thorough assumptions. I confirm or clarify your
assumptions after your words:

If my stated assumptions are incorrect, you can
take them as a form of advice. Thanks.

I assume your single-supply circuit is referenced to
a "pseudo ground" developed by that divider.

You are right.

I assume the amplifier supplies are well bypassed
to that pseudo ground at the frequencies you care
about, (and beyond for stability's sake).

The ultrasound frequency is 40kHz. The signal is burst pulses(5 --10
pulses in one burst). There is a 0.1uF cap connected between the
amplifier supply and the real ground (the battery negative end). From
pseudo ground to the real ground, there are one 10uF and one 0.1uF cap.
I'm not sure the concept of bypassing the pseudo ground. Could please
explain it? thanks.

I assume that your amplifiers stages are the simple,
Rfeedback/Rin configuration and nothing really
boneheaded is hidden by the invisible schematic.

Yes. Each stage is a simple Rfeedback/Rin configuration. The positive
input of the amplifiers are directly connected to the pseudo ground.
The stages are coupled by 1000pF caps.

I assume that you would not say "noise" when you
meant "oscillation".

You are right.

I assume that you would not say "noise" when you
mean "interference getting into the input because of
inadequate shielding or shield connection".

you are right.

I assume that you know about frequency selective
filtering to remove noise from those parts of the
spectrum you do not intend to capture as signal.

I didn't do any frequency filtering on the amplifier circuits. Maybe I
should do it. Could you give some suggestion about this?

Now, given all that, you should expect some noise,
and you should expect the most in the last gain stage.
So I do not see what can be done unless the noise
is excessive, and even then I would need to see a
schematic. See http://www.tech-chat.de/aacircuit.html
for a tool enabling you to post your schematic here.

I can't read the language on that website. The schematic is real no
more than I described.

Thanks again

 

[Mark]

Dear friends,

I am making an ultrasound signal receiver. The transmitter and receiver
is apart from each other about 6 meters. The circuit is powered by a 9v
battery. I used three amplifiers from two TL082. The reference voltage
(about 4.5v) is generated from voltage divider (two 100k resisters in
series). All amplifiers are in inverting input mode. First and second
stages are configured as: 10k input resister and 500k feedback
resister. The signal output in the second stage is very good. However,
the signal from the third stage (input res.= 10k, output res. = 200k)
is significantly corrupted by noise. Please give me some advice to
clean up the amplified signal.

Thanks,

Zhi Yang

How big is the signal at the output of the 3rd stage?

Is it clipping?

Is it random noise you have or an oscillation?

You may need a small pull up or pull down on the output of the 3rd
stage.

Mark

 

[for_idea]

Dear Larry,

Here are your two other comments. My words are after yours:

I assume that the choice of TL082 is appropriate
for the source impedance of your transducer. If
you are not sure of this, you should mention what
it is and what you know about that impedance.

I didn't really take impedance matching into consideration when I made
the circuit. The only information about ultrasound receiver is:
Part number: 255-400ER18
Center Freq.:40kHz+/-1kHz
Bandwith: 1.5kHz(-6dB)
Capacitance@1kHz: 2600pF+/-20%
Allowable input power: 0.2W
input voltage:20v max

I assume that your choice of a 10k input resistor
is made with knowledge of thermal noise, and is
not the source of input referred noise higher than
your application can tolerate. This again depends
on the transducer characteristics.

I understand the higher the resistance I use, the more thermal noise in
circuit, but if I use small resistance, the circuit will consume a lot
of power. 10k seems at the trade-off point.

Thanks again

 

[Larry Brasfield]

Dear Larry,

Thanks for your thorough assumptions.

They were not as thorough as I later thought
appropriate. See my later post about this.

....

You are right.

The ultrasound frequency is 40kHz. The signal is burst pulses(5 --10
pulses in one burst). There is a 0.1uF cap connected between the
amplifier supply and the real ground (the battery negative end). From
pseudo ground to the real ground, there are one 10uF and one 0.1uF cap.
I'm not sure the concept of bypassing the pseudo ground. Could please
explain it? thanks.

You appear to have done it without knowing its name.

Yes. Each stage is a simple Rfeedback/Rin configuration. The positive
input of the amplifiers are directly connected to the pseudo ground.
The stages are coupled by 1000pF caps.

Good. Depending on your signal bandwidth
(likely to be narrow thru air), those caps are
about right for the high pass portion of the
filtering you should have.

["noise" not oscillation or interference]

I didn't do any frequency filtering on the amplifier circuits. Maybe I
should do it. Could you give some suggestion about this?

With some caps across your feeback resistors,
and ajustment of your interstage coupling caps
to shave off the low end, you can get most of
the noise suppressed that lies outside your signal
band. With all those op-amps, you could build a
better filter, but I hesitate to go there until we see
what kind of detection you are doing. Its selectivity
may well obviate the need for a lot of care in the
gain stages.

I should have mentioned that a measure of the
noise you do see, (together with the bandwidth
of the o'scope if that is your measuring tool),
would be necessary to determine whether it
is excessive or to be expected.

I can't read the language on that website. The schematic is real no
more than I described.

You do English fairly passably. Go here:
http://www.tech-chat.de/download.html
and pull down
http://www.tech-chat.de/files/AACircuit128beta.zip
and do the obvious things. For stuff like this, some
diagrams can save thousands of words.

Thanks again

You are most welcome.

 

[Larry Brasfield]

Dear Larry,

Here are your two other comments. My words are after yours:

That is normal Usenet convention. No need to say it.
(And some complain when anything else shows up!)

I didn't really take impedance matching into consideration when I made
the circuit. The only information about ultrasound receiver is:
Part number: 255-400ER18
Center Freq.:40kHz+/-1kHz
Bandwith: 1.5kHz(-6dB)
Capacitance@1kHz: 2600pF+/-20%
Allowable input power: 0.2W
input voltage:20v max

Do you have any reasonable way of measuring the
impedance, especially the real part of it? From a
guestimate based on the capacitance, I think your
10K input resistor may be adding several dB to
the noise that comes from the transducer.

I understand the higher the resistance I use, the more thermal noise in
circuit, but if I use small resistance, the circuit will consume a lot
of power. 10k seems at the trade-off point.

Have you considered that the impedance
level of the input/feeback network only
affects the power consumption while
largish signals are present, and does
not much affect quiescent power?

I would have to say that if you do not
know the source impedance, it is hard
to make the tradeoff you mention. And
that tradeoff should be made differently
for the input stage than the rest.

Thanks again

You're still welcome.

 

[Tim Wescott]

Hello John,



He said 40kHz. But you are right, 50 times is pushing it for a 3MHz
unity gain BW opamp. And per datasheet that's only the typical value.
Quite frankly, for a 40kHz ultrasound amp I'd do it with plain transistors.

Joerg, you do _everything_ with plain transistors.

 

[Joerg]

Hello Tim,

Joerg, you do _everything_ with plain transistors.

Not everything, only when it is cheaper. At two cents a pop these things
are a bargain. But, I did use a 14106 recently. As a current controlled
oscillator...

Seriously, in this case it just might make sense to do it with
discretes. When you need lots of gain at frequencies above the audio
range things get to be expensive with chips.

Regards, Joerg

 

[Larry Brasfield]

[John Woodgate once wrote:]

He said 40kHz. But you are right, 50 times is pushing it for a 3MHz unity gain BW opamp. And per datasheet that's only the typical
value. Quite frankly, for a 40kHz ultrasound amp I'd do it with plain transistors.

Considering that the circuit is battery powered, the
op-amp approach can make a lot of sense. And
unless the actual closed loop gain needs to be
stable and predictable, running at a nominal loop
gain of 1.5 (as the OP's circuit does) may well be
just fine. The sort of device he is conjuring tends
to be narrowband without any particular phase
response requirement or sensitivity to the distortion
that comes up with low loop gain.

That said, the OP may wish to elaborate upon his
need, or lack thereof, for precise amplification.

 

[for_idea]

Dear John,

Thanks for offerring ideas. My answers follow your words below.

(in <[email protected]>) about
'amplify 40kHz audio signal using TL082: first two stages are fine, but
high noise from the third stage', on Wed, 16 Mar 2005:

For what frequency?

40kHz ultrasound burst

50 times gain from a TL082 at an'ultrasonic' frequency is pushing its
capability. It MAY be OK, but a faster op-amp would be better.

As I observed, for first two stages, the amplifier works very well.
Only problem is at the third stage. But choosing a faster op-amp is a
good idea.

What is the peak-to-peak amplitude? With +/- 4.5 V supplies, you can
only get about 8 V peak-to-peak.

It's ok for signal get satiated since I just want to know the time
position of the signal. If the signal is contaminated by significant
noise, the follow-up envelop detection circuit will produce bad
waveforms.

How are you coupling the stages together? If you have DC-coupled them
you are amplifying the first op-amp's input offset by 50 x 50 x 20 = a
lot. 100 uV of offset at the input becomes 5 V at the output - enough to
seriously compromise your results.

The stages are coupled with 1000pF caps.

--
Regards, John Woodgate, OOO - Own Opinions Only.
The good news is that nothing is compulsory.
The bad news is that everything is prohibited.
http://www.jmwa.demon.co.uk Also see http://www.isce.org.uk

Thanks again
Zhi

 

[Joerg]

Hello Larry,

Considering that the circuit is battery powered, the
op-amp approach can make a lot of sense. ...

Could be but the fairly old TL082 draw around 1.5mA per amp so that
would be 4.5mA for three of them. Probably 6mA because he uses duals.

A typical 9V battery would be gone in around 20 hours.

Regards, Joerg

 

[Joerg]

Hello Luhan,

Hey, how about linear-biased cmos inverters? Lots of bandwidth, only
one resitor and one capacitor per stage. Comes in handy Six-Packs.

Like a 74HCU04? Or a similar CD series? You'd have to stay away from
VCC/2 though to avoid lots of quiescent current.

Regards, Joerg

 

[Luhan Monat]

How big is the signal at the output of the 3rd stage?

Is it clipping?

Is it random noise you have or an oscillation?

You may need a small pull up or pull down on the output of the 3rd
stage.

Mark

Hey, how about linear-biased cmos inverters? Lots of bandwidth, only
one resitor and one capacitor per stage. Comes in handy Six-Packs.

 

[Larry Brasfield]

Hello Larry, Hi, Joerg.

Could be but the fairly old TL082 draw around 1.5mA per amp so that would be 4.5mA for three of them. Probably 6mA because he uses
duals.

A typical 9V battery would be gone in around 20 hours.

Good point, and a good reason to choose a different part.
There are better devices with respect to GBW per mA
of supply current. For example, the LT1492 is about 5
times better with comparable noise performance.

There may be cost issues at work in the OP's choice,
and he may be working on something that is on only
a fraction of the time, so that old workhorse may be
a good choice. Otherwise, it clearly is not.

I look at your discrete design advocacy with some
interest, having often been in that camp myself. But
there has lately (well, for some years!) been a fact
that changes those equations. It used to be that
component capacitances were smaller than or
comparable to what leads and traces demand.
Now, with much smaller devices available within
the internal circuitry of IC's, they can deliver
speed versus power that is challenging (to say
the least) to achieve with discrete circuits, even
when fly-turd passives and SOT-23's are used.

 

[Larry Brasfield]

[Correcting a sense error in place.]

 

[Joerg]

Hello Larry,

... Now, with much smaller devices available within
the internal circuitry of IC's, they can deliver
speed versus power that is challenging (to say
the least) to achieve with discrete circuits, even
when fly-turd passives and SOT-23's are used.

So far I haven't seen that advantage that much. Very fast chips are
usually very expensive, meaning they are out of budget for many
projects. On the discrete front there are lots of blazingly fast SiGe
transistors. Even some really old regular ones can still run circles
around chips. An example is ye olde RF work horse, the BFS17A. It can
ramp several milliamps in under a nanosecond.

But there is often a case for chips. I must admit that I have used the
uA733 a lot in the old days. That was because volumes of it were built
into VCRs so it became a cheap device. Whenever I use an IC I try my
utmost to avoid a single sourced part. This is because I have seen too
many people including clients get zinged by that. Heck, I even flung a
SMPS design to semi-discrete once and even after a decade in production
no PWM chip has come close in cost.

In ultrasound apps I'd say that all but two of my front-end designs were
discrete. The two that weren't used the AD603 which is a marvelous chip
but at around $5 it is too expensive for some designs.

Regards, Joerg